Powered by Zoomin Software. For more details please contactZoomin

MarkLogic MCP Server

Glossary

  • Last Updated: September 9, 2026
  • 1 minute read
    • Documentation

Shared terminology used across the MarkLogic MCP Server and Retrieval API documentation.

Term Meaning
Hyperforge The agent framework used to build and run the MCP server in this repository. The Python package name is hyperforge.
MCP Model Context Protocol. The interface used by clients to call server tools.
MCP tool A callable action exposed by the server, such as Retrieve, RetrieveDefinition, or Augment.
MarkLogic Retrieval API (MRA) The REST API that provides search, definition, and document augmentation endpoints.
Retrieve The /v1/retrieve endpoint and corresponding tool for hybrid search across text, labels, filters, and vectors.
Retrieve definition The /v1/retrieve/definition endpoint and corresponding tool that returns available labels, filters, vector metadata, and related configuration.
Augment The /v1/retrieve/augment endpoint and corresponding tool that returns full document content by URI.
Label A MarkLogic collection name used for retrieval filtering. In request payloads, labels are typically mapped to MustHave or MustNotHave.
Filter A constraint-based search predicate used to narrow retrieval results by property, range, or view join.
Vector metadata Configuration describing available vector embeddings, including the TDE schema, view, vector column, model name, and dimensions.
Processing metadata Customer-defined metadata extracted from document content using processingMetadataPath.
Provenance Response metadata that identifies the source document and XPath for a retrieved chunk.
Chunk A fragment of a larger document that can be retrieved and cited independently.
Chunk-level retrieval Augmenting a specific document fragment by passing a {uri, xpath} object to /v1/retrieve/augment.
TDE Template Driven Extraction. MarkLogic's framework for exposing document content as relational views.
Optic MarkLogic's query and analytics API used for view-based joins and vector-backed retrieval.
Alert